No project description provided
Project description
Graph Matching
Models Used
-
SimGNN:
- Encoder:
- Inputs: Initial one-hot encoded node embedding matrix $U \in R^{NXD}$
- Outputs: Aggregated node Embedding Matrix $U \in R^{NXD}$
- Uses: Neighbour Aggregation with Conv Nets (SAGE, GCN, GAT)
- Attention Mechanism:
- Inputs: Node Embedding Matrix $U \in R^{NXD}$
- Outputs: Attention Weighted Graph Embedding Vector $h \in R^{D}$
- Uses: Non linear weighted transform ($\tanh$) for context, sigmoid layers for att. weights, $\sum$ aggregate for h
- Graph Interaction Extraction:
- Inputs: Graph Embedding Vectors $h_{q}, h_{c} \in R^{D}$
- Outputs: Interaction Score Vector $g \in R^{K}$, K being the depth of the NTN
- Uses: Neural Tensor Network
- Score Predictor:
- Inputs: Graph Similarity Score Vector $g \in R^{K}$
- Outputs: Graph Similarity Score s
- Uses: Fully Connected Network
- Encoder:
-
GMN Embed:
- Encoder:
- Inputs:
- Initial Node Representation Matrix $U \in R^{NXD}$
- Initial Edge Representation Matrix $X \in R^{NXN}$
- Outputs: Encoded Node and Edge Embedding Vectors $H^{0} \in R^{NXD}$ and $E \in R^{NXN}$
- Uses: Multi Layer Perceptron Networks
- Inputs:
- Propagation:
- Inputs: Encoded embeddings $H^{0} \in R^{NXD}$ and $E \in R^{NXN}$
- Encoder:
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Close
Hashes for graph_retrieval_toolkit-0.0.1.tar.gz
Algorithm | Hash digest | |
---|---|---|
SHA256 | 29c3af0c28fb724f64e63314c07ae0aa7840b1673907e3fe66ee0c31cb71c323 |
|
MD5 | df2efcaa8ce11941f61e7a8a836bfc5f |
|
BLAKE2b-256 | dcb0788f7f78e0a7592a8357c326c16d17511c806babbffc0261866377a60612 |
Close
Hashes for graph_retrieval_toolkit-0.0.1-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 39e1b4ca2dcb82b9633f9db8cb39900807984c9c99866b4e9964cb725a68e44d |
|
MD5 | 1ca22f3ac70a2677af4d20811cd07e1a |
|
BLAKE2b-256 | 1c5dad3844dc5401640de86f3db40cb0988d65ffef3ec12c3c80bb6ec4d70a05 |